diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -2120,6 +2120,15 @@
   llvm::Type *Ty = Src->getType();
   ShiftAmt = Builder.CreateIntCast(ShiftAmt, Ty, false);
 
+  // For WebAssembly we want to generate the target-specific rotate builtin, 
+  // rather than generating fshl/fshr intrinsics.
+  if (getTarget().getTriple().isWasm()) {
+    unsigned IID = IsRotateRight ? Intrinsic::wasm_rotr_i32
+                                 : Intrinsic::wasm_rotl_i32;
+    llvm::Function *F = CGM.getIntrinsic(IID);
+    return RValue::get(Builder.CreateCall(F, { Src, ShiftAmt }));
+  }
+
   // Rotate is a special case of LLVM funnel shift - 1st 2 args are the same.
   unsigned IID = IsRotateRight ? Intrinsic::fshr : Intrinsic::fshl;
   Function *F = CGM.getIntrinsic(IID, Ty);
diff --git a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
--- a/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
+++ b/llvm/include/llvm/IR/IntrinsicsWebAssembly.td
@@ -341,4 +341,16 @@
                         [],
                         [IntrReadMem]>;
 
+//===----------------------------------------------------------------------===//
+// Rotate Intrinsics
+// These are lowered from the target independent intrinsics to avoid
+// funnel shift optimizations
+//===----------------------------------------------------------------------===//
+
+def int_wasm_rotl_i32 : 
+  DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+
+def int_wasm_rotr_i32 : 
+  DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+
 } // TargetPrefix = "wasm"
diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyInstrInteger.td b/llvm/lib/Target/WebAssembly/WebAssemblyInstrInteger.td
--- a/llvm/lib/Target/WebAssembly/WebAssemblyInstrInteger.td
+++ b/llvm/lib/Target/WebAssembly/WebAssemblyInstrInteger.td
@@ -107,6 +107,12 @@
 def : Pat<(rotl I64:$lhs, (and I64:$rhs, 63)), (ROTL_I64 I64:$lhs, I64:$rhs)>;
 def : Pat<(rotr I64:$lhs, (and I64:$rhs, 63)), (ROTR_I64 I64:$lhs, I64:$rhs)>;
 
+// Lower the rotate intrinsic to a rotate instruction
+def : Pat<(int_wasm_rotl_i32 I32:$lhs, I32:$rhs),
+          (ROTL_I32 I32:$lhs, I32:$rhs)>;
+def : Pat<(int_wasm_rotr_i32 I32:$lhs, I32:$rhs),
+          (ROTR_I32 I32:$lhs, I32:$rhs)>;        
+
 defm SELECT_I32 : I<(outs I32:$dst), (ins I32:$lhs, I32:$rhs, I32:$cond),
                     (outs), (ins),
                     [(set I32:$dst, (select I32:$cond, I32:$lhs, I32:$rhs))],